Docker Deployment¶

This guide shows you how to deploy the FIRST Inference Gateway using Docker and Docker Compose.

Prerequisites¶

Docker Desktop 4.29+ (or Docker Engine 24+) with Docker Compose v2
Git
Globus Account and registered applications
At least 4GB RAM available for containers

Step 1: Clone the Repository¶

git clone https://github.com/auroraGPT-ANL/inference-gateway.git
cd inference-gateway

Step 2: Configure Environment¶

Create a .env file from the example environment file and customize the .env file following the instructions found in the example file:

cp env.example .env

Make sure you include all of the Globus UUIDs and secrets generated during the Globus setup stage. You can generate the SECRET_KEY variable with the following Django command (if installed):

python -c 'from django.core.management.utils import get_random_secret_key; print(get_random_secret_key())'

Production Security

For production deployments:

Set RUNNING_AUTOMATED_TEST_SUITE=False
Set DEBUG=False
Use secure passwords and secrets
Add your domain to ALLOWED_HOSTS or use "*" if appropriate
Add at least one Globus High Assurance policy (GLOBUS_POLICIES)
Set authorized IDP domains (AUTHORIZED_IDP_DOMAINS) to match the policy
Consider using secrets management (e.g., Docker secrets)

Step 3: Start the Services¶

cd deploy/docker
docker-compose up -d --build

The docker-compose.yml includes:

Core Services¶

inference-gateway: Django API application (internal port 8000)
postgres: PostgreSQL 15 database (internal port 5432)
redis: Redis 7 cache (internal port 6379)
nginx: Reverse proxy (internal port 80 exposed to localhost port 8000)

Optional Services¶

You can add these to your compose file:

prometheus: Metrics collection
grafana: Visualization dashboard

Verify that the core-service containers are running:

docker-compose ps

Step 4: Initialize the Database¶

Run migrations:

docker-compose exec inference-gateway python manage.py makemigrations
docker-compose exec inference-gateway python manage.py migrate

Step 5: Test the Gateway¶

Check that the gateway is running:

curl http://localhost:8000/resource_server/whoami

If everything is running, the command should give you the following error:

Missing ('Authorization': 'Bearer <your-access-token>') in request headers.

Common Commands¶

View logs¶

# All services
docker-compose logs -f

# Specific service
docker-compose logs -f inference-gateway

Restart services¶

# All services
docker-compose restart

# Specific service
docker-compose restart inference-gateway

Stop services¶

docker-compose down

Stop and remove volumes (clean slate)¶

docker-compose down -v

Access container shell¶

docker-compose exec inference-gateway /bin/bash

Run Django management commands¶

docker-compose exec inference-gateway python manage.py <command>

Updating the Deployment¶

Pull latest changes:

git pull origin main
docker-compose build
docker-compose up -d
docker-compose exec inference-gateway python manage.py migrate

Troubleshooting¶

Gateway won't start¶

Check logs:

docker-compose logs inference-gateway

Common issues:

Missing environment variables
Database connection failed
Port 8000 already in use

Database connection errors¶

Verify PostgreSQL is running:

docker-compose ps postgres

Check database logs:

docker-compose logs postgres

Can't access admin panel¶

Ensure you created a superuser:

docker-compose exec inference-gateway python manage.py createsuperuser

502 Bad Gateway from Nginx¶

Vefiry that the gateway container is running:

docker-compose ps inference-gateway

Verify nginx configuration:

docker-compose exec nginx nginx -t